[Refactor] [10/N] to simplify the vLLM openai completion serving architecture (#32369)

Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
This commit is contained in:
Chauncey
2026-01-15 07:41:34 +00:00
committed by GitHub
parent ae1eba6a9a
commit 4c1c501a7e
43 changed files with 777 additions and 642 deletions
+5 -3
View File
@@ -31,11 +31,13 @@ if TYPE_CHECKING:
from vllm.entrypoints.openai.chat_completion.protocol import (
ChatCompletionRequest,
)
from vllm.entrypoints.openai.engine.protocol import (
from vllm.entrypoints.openai.completion.protocol import (
CompletionRequest,
)
from vllm.entrypoints.openai.engine.protocol import (
StreamOptions,
)
from vllm.entrypoints.openai.serving_models import LoRAModulePath
from vllm.entrypoints.openai.models.protocol import LoRAModulePath
else:
ChatCompletionRequest = object
CompletionRequest = object
@@ -281,7 +283,7 @@ def should_include_usage(
def process_lora_modules(
args_lora_modules: list[LoRAModulePath], default_mm_loras: dict[str, str] | None
) -> list[LoRAModulePath]:
from vllm.entrypoints.openai.serving_models import LoRAModulePath
from vllm.entrypoints.openai.models.serving import LoRAModulePath
lora_modules = args_lora_modules
if default_mm_loras: